Paul Gilmartin wrote:

>IBM 7030 did something like the latter, but neither is right; it's worse.

>... Note that "CAMEL" comes after "Camel" but before "canary".  So you can't 
>simply say either that A<a or A>a.


Clark Morris wrote: 

>A more important reason was that addition of mixed case in either ISO/ASCII 
>changed a compare from a simple Compare Logical Character into a subroutine.  
>While this was always true if a true dictionary or phone book sort is wanted, 
>this would make it true for virtually all compares.  Should A = a?  If not 
>should the sequence be A,a,B,b ....?

This is a common and tough problem. Term for that is Collation. You decide what 
sequence do you want to use (and standardize on).

You select ISO, ASCII, EBCDIC or other set or alphaset in whatever language and 
then you decide what character comes before what other. Do you want to sort 
from top to bottom or left to right depending on language?

Another sequence (collation) problem which is also sometimes overlooked: one 
example - 'North' certainly comes before 'Northern'. This can be resolved by 
giving a space higher priority for this case.

For more background info: https://en.wikipedia.org/wiki/Collation

For DFSORT this:

https://www.ibm.com/support/knowledgecenter/SSLTBW_2.2.0/com.ibm.zos.v2r2.icea100/cultenv.htm
 
https://www.ibm.com/support/knowledgecenter/SSLTBW_2.2.0/com.ibm.zos.v2r2.icea100/stand.htm

UNICODE in IBM: 

https://www.ibm.com/support/knowledgecenter/SSLTBW_2.3.0/com.ibm.zos.v2r3.cunu100/collati.htm

I know DB2 have also issues with collation in data and tables, but I don't have 
now access to a DB2 system to check it out.

For z/OS Unix things, I believe these have also its own collation setup and 
defaults. 

That is hard to sort it out... ;-)

Groete / Greetings
Elardus Engelbrecht

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to